Joint scheduling method based on urban public transport resources

Abstract

The invention provides a joint scheduling method based on urban public transport resources, which focuses on a bus system and a shared bicycle system, uses a reinforcement learning long-term optimal scheduling and cooperative scheduling strategy on the basis of time-space prediction, can achieve cooperative scheduling between the bus system and the shared bicycle system, solves the possible localgreedy problem, can schedule other traffic resources can be dispatched in time to meet travel requirements of users when a certain traffic service is temporarily unavailable or inapplicable. The method comprises the following steps: according to recorded crowd flow data at different time and places and people flow changes borne by various vehicles, pre-constructing a time-varying demand flow graphfor people to take the vehicles; and then, taking the current station state and the future predicted flow diagram as the state of the current system, and performing collaborative and efficient scheduling on the current multiple traffic systems by utilizing a reinforcement learning technology.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a reinforcement learning method with a people flow prediction method and a joint scheduling of traffic resources. Background technique [0002] In recent years, the problem of traffic congestion in modern cities has increasingly become a problem for residents. As shown in Baidu's traffic report, Beijing's commute stress index reached an astonishing 1.973 during peak hours, resulting in longer travel times and more vehicle queues. Previous studies have shown that through rational scheduling, such as rescheduling bike-sharing systems and optimizing bus transportation systems, traffic efficiency can be significantly improved without consuming excess resources. [0003] The inventors found that there are still two shortcomings that limit the performance of the dispatching system: (1) only a single dispatch in a short period of time is considered, while ignoring the redistribut...

AI Technical Summary

Problems solved by technology

[0003] Find through research of the present inventor, still have two shortcomings to limit the performance of dispatching system: (1) only consider the single dispatching in short time, and ignore traffic resource redistribution phenomenon after first traffic dispatching; (2) current dispatching system Only focus on one type of traffic scheduling

The multimodal nature of urban public transport is largely underutilized

Images